Defaults are half PPR, 12 teams, one QB, two RB, two WR, one TE, one flex, playoffs weeks 15 to 17. Change anything; the page remembers it. Or import a Sleeper league and it sets itself.
Loads the league's scoring, roster slots and size, your roster, and whichever team you are trading with.
Type names, pick from the list. Kickers and defenses count as nothing over replacement in redraft, so they are not on the board.
The number. TwoGunGPT's own rest-of-season projection for every player, every remaining week: his team's environment (plays, pass-rush split, expected points, off this week's line and a points model after that), times his role (his share of the team's targets and carries: last season's share as the prior, this season's games updating it with the recent weeks heaviest, snap share as the depth chart), times his efficiency (shrunk hard toward position norms), times his chance of playing (ESPN's injury report, the box-score record when it says nothing). Byes are zero. Public data only: box scores and snap counts, the schedule, the lines, the injury report. No preseason projections anywhere. No third-party projections.
The value. A player is worth what he scores over the worst starter at his position in a league your size, summed week by week, never below zero in a week, with your playoff weeks weighted by the win-now knob. The side taking fewer players gets the roster spot it opens: a bench-level player about one week in seven. Loaded a roster? The trade is scored by the change in your best starting lineup, week by week, before and after.
The scorecard. Below: the same engine run as-of weeks 4, 8 and 12 of the last two seasons, ranked against what happened, next to two dumb baselines. Read it before you trust it.
Type what you give and what you get, set your league's scoring and size, and TwoGunGPT scores both sides on its own rest-of-season projections: points over the worst starter at each position, week by week, with your fantasy playoff weeks weighted the way you tell it to. You get a verdict (coin flip, slight edge, win, fleece), the numbers behind it, and what would make it fair.
A player is worth what he scores ABOVE the worst starter at his position in a league your size, summed over the weeks left, never below zero in a week, with a bye week worth nothing. The side that takes fewer players gets a small credit for the roster spot it opens (a bench player plays about one week in seven). If you load your roster, the page scores the trade by the change in your best starting lineup instead, which is the only question that matters in redraft.
They are TwoGunGPT's own rest-of-season projections, rebuilt every Tuesday from public data: box scores and snap counts (nflverse), the schedule and this week's betting lines, and ESPN's injury report. A player's week is his team's environment times his share of its targets and carries times his efficiency, each projected on its own footing and shrunk toward what is stable. No third-party projections are used. No preseason projections are used anywhere.
Because they are someone's August opinion, and by October the tape is better evidence. The prior here is last season's ROLE (his share of his team's work), which this season's games update every week with the recent weeks weighted heaviest. Efficiency is shrunk hard because most hot starts are noise.
The scorecard on this page is the test: the same engine run as-of weeks 4, 8 and 12 of past seasons against what players actually scored the rest of the way, next to two dumb baselines (season-to-date points a game and the last three weeks). It ships only because it beats both at every position; the table shows the one cell where it did not on the holdout year. Rest-of-season projections have a ceiling, so the verdict says coin flip when it is one.
By points over replacement, not raw points, so two bench pieces do not add up to a star, and the side taking fewer players gets the value of the roster spot it opens. In roster mode the page just re-runs your starting lineup with and without the trade, which handles it exactly.
Yes, from Sleeper: type your Sleeper username, pick the league, and the page loads its scoring, roster slots, your roster and the other team's. Sleeper's league data is public, no login needed. Any league: paste your roster, one name per line.
Yes: twogungpt.com/trade-analyzer/?give=First Player,Second Player&get=Third Player&rec=0.5&teams=12 opens the page with that trade already analyzed.